How to check whether your sales team keeps up with incoming leads

Speed matters, but the best-known numbers are 15 years old

And the work measures a conversation, not a sale

The most cited source is a March 2011 Harvard Business Review piece by James Oldroyd, Kristina McElheran and David Elkington. They submitted a test lead through the websites of 2,241 US companies and timed the responses. 37% responded within an hour, 16% within one to 24 hours, 24% took longer, and 23% never responded. Among those that responded within 30 days, the average response time was 42 hours.

A separate analysis covered 1.25 million leads at 29 consumer and 13 business-to-business companies. Firms that tried to contact a lead within an hour were nearly 7 times as likely to qualify it, meaning a meaningful conversation with a key decision maker, as those that tried an hour later, and more than 60 times as likely as those that waited 24 hours or more.

Three limits apply. The data is 15 years old and predates messaging apps. The second comparison is observational: faster firms may differ in other ways, such as their lead mix. And the work measures a conversation, not a sale. A widely repeated 2007 study about “the first five minutes” is left out: we could not find its original report with a method.

Measure minutes from lead to first attempt, not “we did call back”

You need two timestamps per record: lead arrival and the first attempt to reach the person by call, message or email. An automatic “we received your request” reply is not an attempt. Count after-hours leads separately: mixing them with daytime leads hides both problems.

Do not rely on the average. Illustrative example: nine leads got a first attempt after 10 minutes, and a tenth, sent on Friday evening, after 16 hours. The average is 105 minutes; the median, the middle value of the sorted list, is 10 minutes. The risk shows in a third number: 1 in 10 leads waited over an hour. Report the median and the share beyond your threshold.

Each CRM report starts the clock differently. HubSpot’s lead response time runs from assignment to a user until that user’s first interaction, including marking a task as in progress or complete; reassignment resets it. Kommo’s guide points to unanswered conversations, median reply time and the longest wait. Check on three records where the clock starts and stops.

One missed call does not mean the customer is not interested

The second question: how many attempts, in which channels. The person may ignore unknown numbers or expect a reply in the app they used. One unanswered call says nothing about their interest.

There is no universal number of attempts: none of the sources we found sets one with a stated method. Below is an editorial example; adapt it to your cycle. Agree the sequence in advance, log every attempt, and set “could not reach” only after the last step.

CRMs support this. A HubSpot sequence sends timed emails and follow-up tasks and unenrolls a contact who replies or books a meeting. Kommo tasks have an owner and a deadline.

1. Immediately: call, and if there is no answer, send a WhatsApp message or text with the salesperson’s name and reason.
2. One to two hours later: a second call.
3. Next business day: a call at another time of day, plus an email if you have the address.
4. Third business day: a final message asking whether and where to continue.
5. After the last step: status “could not reach” with the number of attempts.

Statuses must separate “could not reach” from “did not need it”

When one loss reason covers everything, such as “not relevant”, the audit becomes guesswork. Split reasons into three groups: follow-up (not reached, no callback, no proposal), fit (wrong product, location or budget) and customer decision (chose a competitor, postponed). Defining fit is covered in our piece on lead quality.

A new Bitrix24 pipeline has eight default stages, including Deal lost and Analyze failure, and the help centre notes that multiple failed stages help analyse why deals were lost. HubSpot deals have a standard Closed lost reason property. Kommo’s guide counts missing Lost reasons and tasks as signs of incomplete data, and warns that Lost leads are attributed to the currently responsible user, not necessarily whoever worked the lead.

A worked audit of 50 leads: almost half were lost to follow-up

Follow-up losses clearly outnumber leads that do not fit

An illustrative example with invented numbers. A company took 50 leads from one month, each at least 30 days old, and filled in an audit sheet for each. The age rule stops you counting a lead still in progress as lost; see our piece on honest lead-to-sale conversion.

The result: 14 were never reached, 9 agreed to a callback that never happened, 10 were confirmed as not a fit, 11 are “thinking about it”, and 6 became deals. Total: 50.

Of the 14 never reached, 8 had fewer than three attempts in one channel. Of the 11 “thinking about it”, 7 had no task for a next step. With the 9 missed callbacks, that is 24 of 50 leads, or 48%, lost to follow-up. Confirmed poor fit: 10, or 20%.

A sample this small is imprecise. The Wilson interval, a standard way to estimate the range for a proportion, gives roughly 35% to 61% at 95% confidence for 24 of 50, and 11% to 33% for 10 of 50. Enough for a first decision: follow-up losses clearly outnumber leads that do not fit. Too small to compare channels or people.

Check recordings and messages on a sample from every status

A status is the salesperson’s opinion; only the conversation confirms it. Take two or three recordings or threads from each final status, picking records by list position, say every fifth, rather than loud cases from memory. You will see typical work and labelling errors.

Ask four questions. Did the salesperson find out what the customer needs? Did they propose a dated next step? Does the status match what was said? Is the next step logged as a task? If “not a fit” on the record sounds like “call me next month” in the recording, that is a follow-up failure.

In Kommo, recordings from a connected telephony integration can be played and downloaded from the call log. If many conversations are missing from records, fix tracking first, or the audit will measure data gaps.

A lead handling policy fits on one page

After the audit, write rules that close the losses you found, each with a deadline and a place in the CRM where compliance is visible. The HBR authors blame slow responses partly on pulling leads from the CRM daily rather than continuously and on distributing them by geography and “fairness”. So start with how a lead reaches a person.

1. A lead gets an owner the moment it arrives, automatically or via an on-duty rota.
2. First-attempt deadline in business hours is set in minutes; after-hours leads have their own rule.
3. The attempt sequence is set by channel and day, and every attempt is logged.
4. A record can be closed only with a reason from the agreed list.
5. Every open record has a dated next-step task.
6. Weekly, the sales manager reviews median first-attempt time, the share beyond the threshold and records without tasks.

Team assignment for the next two weeks

For the sales manager and the CRM owner. Output: a 50-lead audit table, a short report with counts and a draft policy. The procedure is the editors’ suggestion, not from CRM documentation.

• Select 50 leads from one month, each at least 30 days old, by order of arrival, and fill in an audit sheet for each.
• Calculate median time to first attempt and the share beyond the threshold, separately for after hours.
• Group outcomes into follow-up, fit and customer decision, with counts, and check two or three conversations from each status.
• Propose three policy rules against the largest losses and a date for a repeat audit.

When speed is not the main issue, and what to check first

The audit shows where leads are lost, not why. A fast reply will not save a lead if the ads attract people with a different need or budget. If confirmed poor-fit leads outnumber follow-up losses, fix quality criteria, the offer and the form first.

The HBR data is old and observational, so do not treat “7 times” as a forecast. Compare outcomes for your own leads first contacted before and after your threshold, over the same observation period. Leads “thinking about it” may buy later; with a long sales cycle, repeat the audit a month on.

Recordings contain personal data: limit access to reviewers and check local rules with a lawyer. This publication is produced with the involvement of WEBNOSTRA, which builds sales teams and runs such audits; the service link is below, but your team can run this check itself.

Key decisions

1. measure time to first attempt from CRM timestamps, separately for business and after hours
2. do not close a lead as unresponsive until a pre-agreed sequence of attempts is complete
3. check statuses and loss reasons against recordings or messages on a sample

What to discuss with your team

• Every lead has arrival and first-attempt timestamps
• “could not reach” is set only after the full sequence
• Loss reasons are split into follow-up, fit and customer decision
• Statuses are checked against recordings and messages on a sample
• Every open record has a dated task

Sources

1. Oldroyd J., McElheran K., Elkington D. The Short Life of Online Sales Leads. Harvard Business Review, March 2011hbr.org · checked 31/10/2026
2. HubSpot Knowledge Base: Create sales reports in the sales analytics suiteknowledge.hubspot.com · checked 31/10/2026
3. HubSpot Knowledge Base: HubSpot's default deal propertiesknowledge.hubspot.com · checked 31/10/2026
4. HubSpot Knowledge Base: Create and edit sequencesknowledge.hubspot.com · checked 31/10/2026
5. Kommo: Track your win-loss statisticssupport.kommo.com · checked 31/10/2026
6. Kommo: Create and assign taskssupport.kommo.com · checked 31/10/2026
7. Kommo: Analyze calling activitysupport.kommo.com · checked 31/10/2026
8. Kommo: Review your team's sales performancesupport.kommo.com · checked 31/10/2026
9. Bitrix24 Helpdesk: Sales pipelineshelpdesk.bitrix24.com · checked 31/10/2026
10. Bitrix24 Helpdesk: Statuses and dropdowns in CRMhelpdesk.bitrix24.com · checked 31/10/2026
11. NIST/SEMATECH e-Handbook of Statistical Methods: Confidence intervals for proportionsitl.nist.gov · checked 31/10/2026